Find the Original Image Skill (useOSINT): Reverse Search

Makes an agent reverse-search a picture across five engines to find where it came from and who published it first.

Installs: 56.6K; on skills.sh
Stars: 58; on GitHub
Files: 2; besides SKILL.md
Updated: Aug 2026; last repo commit

This is a method skill from useOSINT for reverse image search. It tells the agent to pick engines by what you hold, run at least three, and crop, flip and clean up the image before giving up. The goal is the earliest publication, not any match. Yandex suits faces and non-Western street scenes, Google Lens objects and text, Bing a boxed region, TinEye oldest copies and Baidu Chinese content.

The skill bundles no scripts. Its reference files show local commands such as exiftool, ImageMagick, tesseract, ffmpeg and yt-dlp. It also covers face-search engines and says to use them only with authorization or a protective purpose, because a facial template is biometric data.

Good for

Investigators of romance and investment scams and fake profiles
Media verifiers and disinformation analysts tracing the oldest copy of an image
Brand-protection and insurance claim teams checking for stock photos and reposts

Install find-the-original-image

Claude Code

Copy this prompt.

Install the find-the-original-image skill in Claude Code. Run this command:

/plugin marketplace add useosint/skills
/plugin install useosint@useosint

Then check that it loaded and tell me how to use it.

2
Paste it into Claude Code. It shows the change and asks before running it.

Pick one route. The Claude Code plugin installs all 29 skills of the repo (plugin.json and marketplace.json say 29, the README text says 28, and the repo tree has 29 skill folders). The plugin folder holds only plugin.json and marketplace.json, with no hooks, agents or MCP servers. The first skills CLI line and the folder routes install just this skill; the helper line adds the skills its method names. The README still names the repo useosint/osint-skills; the repo is now useosint/skills and GitHub forwards the old name. install.sh replaces any folder in the target that has the same name as one of the skills (it runs rm -rf on each destination first). Each skill can run on its own, and its sibling skills are called out in its method or its pivot table. There is no Claude app tab because the README documents no upload route and the skill calls for local tools.

Built from the package data we hold. Not yet tested by us. Source.

What's in the skill

When it loads

Reverse image search across Yandex, Google Lens, Bing Visual Search, TinEye and Baidu to find where a picture came from and who published it first. Use when reverse image searching, identifying a photo, face, logo, product, uniform or building, tracing a profile picture or avatar, finding the oldest copy of an image, checking whether a photo is stock or a repost, or reverse-searching a video by keyframes. Applies to romance and investment scam investigation, fake-profile and synthetic-identity detection, disinformation and media verification, counterfeit and brand-infringement work, and insurance claim review. Reference at useosint.com/skills/find-the-original-image.

Sections in SKILL.md

Pick your engine by what you are holding
What each engine is actually doing
Method
Establishing first publication, not just a match
Face-specific engines and when not to use them
Where this goes wrong
Confidence grading
Worked example
Pivots

Key facts

Author: useOSINT (plugin.json lists the author as useOSINT; the repo is useosint/skills)
License: MIT (LICENSE file, README badge and plugin.json)
Installs on skills.sh: 56,558 (Nov 2, 2026)
skills.sh audits: Gen Agent Trust Hub pass, Socket pass, Snyk warn (Nov 2, 2026)
GitHub stars: 58 for the whole useosint/skills repo (Nov 2, 2026)
Last repo commit: Aug 3, 2026
What it includes: 3 files including SKILL.md (counted from the repo tree): SKILL.md, reference/engine-matrix.md and reference/preprocessing.md. It bundles no scripts.
Needs: The image (or video keyframes) and access to the search engines in a browser; optional local tools: exiftool, ImageMagick, tesseract, ffmpeg, yt-dlp
Works in: The README names Cursor and Claude, and says "other agents" can use the skills; the plugin.json and marketplace.json make a Claude Code plugin
Who it is for: Romance and investment scam investigation, fake-profile and synthetic-identity detection, disinformation and media verification, counterfeit and brand work, insurance claim review (SKILL.md description)

What it includes

Chooses Yandex, Lens, Bing, TinEye or Baidu by what you hold
A table maps a face, street scene, product, text, landmark, boxed region, old press photo or Chinese content to the engine that handles it best, and says to run at least three.
Try: “Find where this photo first appeared. Here is the file.”
Read · Pick the engine
Uses the best copy and checks metadata first
Uses the highest-resolution file, not a screenshot, and runs secrets-in-file-metadata on the original before editing copies.
Files · Fix the input
Re-searches crops, mirrored copies and cleaned-up inputs
Crops to distinctive objects, flips horizontally, and for poor inputs upscales, denoises and corrects levels with commands in reference/preprocessing.md.
Runs code · Crop, flip, preprocess
Searches stills taken from a video
Extracts keyframes with ffmpeg and searches frames from the start, middle and end plus any frame with a sign or face.
Runs code · Video keyframes
Treats the oldest date as a ceiling and opens the pages
Sorts TinEye results by oldest, then opens the match pages to read photographer credit, caption, names, date and filename, and checks the archive capture date.
Network · Sort for oldest and read the match
Allows PimEyes-type tools only with a legitimate basis
Names PimEyes, FaceCheck.ID and Search4Faces, says a face hit is unconfirmed on its own, and says not to identify a stranger from a photo.
Read · Face-engine rules
Labels the result confirmed, probable, match only or unconfirmed
Separates a confirmed original source from a match that only shows an image at a URL by a date.
Read · Confidence grading

From the official docs

Permissions

Medium risk

The skill bundles no scripts, but searching with an image shares it with the search engines, and face search carries legal limits. Crop sensitive images, and use face engines only with authorization.

What it can do

Read your data
Create and change things
Delete things
Read-only mode

Watch out for

Uploading an image shares it with the search engine
SKILL.md says uploading a client's image to a commercial engine discloses it to that vendor. For sensitive material, search a crop that omits identifying detail, or do not search. github.com
Face search is biometric data
SKILL.md says a facial template is biometric data under GDPR Article 9, and that Illinois BIPA and Texas CUBI add private claims. Use face engines only with authorization or a protective purpose. github.com
It suggests local commands on your files
reference/preprocessing.md shows commands such as exiftool, magick, tesseract, ffmpeg and yt-dlp. Work on copies, as SKILL.md advises, and read each command first. github.com

Security advisories

No published security advisories as of Nov 2, 2026.

Technical details

Turning on read-only	No read-only mode. The skill bundles no scripts.
Talks to	yandex.com, lens.google.com, bing.com, tineye.com and baidu.com, when the agent or you search with them
Sign-in and keys	No account or key is named in SKILL.md.
Scopes	Reads the image you give it May run local commands from reference/preprocessing.md: exiftool, ImageMagick, tesseract, ffmpeg, yt-dlp Points to uploading the image to Yandex, Google Lens, Bing, TinEye and Baidu

Questions

What does find-the-original-image do?

It gives the agent a method for reverse image search: choose engines by what you hold, run at least three, crop and flip, read the match pages, and grade how sure the result is. It aims at the earliest publication, not any match.

Does it search for me, or do I need accounts?

SKILL.md describes the engines and when to use them and bundles no scripts. Whether your agent can open and upload to Yandex, Google Lens, Bing, TinEye and Baidu depends on its browser tools. No account or API key is named in SKILL.md.

Can I use it to identify a stranger from a face?

SKILL.md says not to. It says face search is for documented authorization or a protective purpose, for example verifying a counterparty in a fraud case, a consenting subject, missing-persons work or your own exposure, and that a face-engine hit alone is unconfirmed.

Is it safe to use?

It bundles no scripts, but its references show local commands run on your image files, and uploading an image to a search engine discloses it to that engine. SKILL.md warns that you may leak a client's image that way. Our advice: search a crop that leaves out identifying detail for sensitive material.

Does it need the other useOSINT skills?

Its method names secrets-in-file-metadata (run on the original first), read-deleted-pages and google-like-a-spy, so install those with it for the full method. The install box has a skills CLI line that adds the three, and the plugin route installs all 29.

How is it different from is-this-photo-real?

This skill finds where a picture came from. is-this-photo-real decides whether it is authentic and correctly captioned, and it hands the earliest-copy search to this one. Both are in the same useOSINT repo.

Information

updated Nov 2, 2026

Repository: useosint/skills
Folder: skills/find-the-original-image

useosint/skills skills.sh

Sources

Docs, issue trackers and discussions this page is built from.

Show all 8

find-the-original-image SKILL.md · github.com
useosint/skills README · github.com
useosint/skills LICENSE · github.com
useosint plugin.json · github.com
useosint marketplace.json · github.com
useosint ETHICS.md · github.com
preprocessing guide · github.com
engine matrix · github.com